Fix "if index"-like test with None test

Two checks related to multidimensional indices were incorrectly apply
generic `if index` tests, which return false if `index` is zero.

This raised errors in two cases:

1. The end-bound of an array a(0), which is bounded by 1, was
   incorrectly set to None rather than 1, and treated as a floating
   index.

2. The `FIndex` iterator would incorrectly terminate higher ranks if the
   current higher rank index was zero.

These errors occur because we use `None` when the index bounds are
unspecified, and are implicitly determined when values are provided.

Both of these tests have been replaced with `if None` tests, which
appear to have resolved these errors.

f90nml/findex.py

@@ -38,7 +38,7 @@ def __next__(self):
         state = self.current[:]
         # Allow the final index to exceed self.end[-1] as a finalisation check
         for rank, idx in enumerate(self.current):
-            if ((not self.end[rank] or idx < (self.end[rank] - 1)) or
+            if ((self.end[rank] is None or idx < (self.end[rank] - 1)) or
                     rank == (len(self.current) - 1)):
                 self.current[rank] = idx + self.step[rank]
                 break

f90nml/parser.py

@@ -669,7 +669,6 @@ def _parse_variable(self, parent, patch_nml=None):
 
         if not v_idx:
             v_values = delist(v_values)
-
         return v_name, v_values
 
     def _parse_indices(self):
@@ -711,7 +710,7 @@ def _parse_index(self, v_name):
                     raise
         elif self.token in (',', ')'):
             # Replace index with single-index range
-            if i_start:
+            if i_start is not None:
                 i_end = 1 + i_start
 
         # Stride index